Implement Addons>ParseTime operator. (#530)

The parse time operator parses an input string according to the
provided format string into a Unix time, the number of seconds /
milliseconds / microseconds / nanoseconds elapsed since January 1,
1970 UTC.

Fixes: #492

Author: helinwang

tensorflow_addons/custom_ops/text/BUILD

@@ -21,3 +21,21 @@ cc_binary(
         "@local_config_tf//:tf_header_lib",
     ],
 )
+
+cc_binary(
+    name = "_parse_time_op.so",
+    srcs = [
+        "cc/kernels/parse_time_kernel.cc",
+        "cc/ops/parse_time_op.cc",
+    ],
+    copts = [
+        "-pthread",
+        "-std=c++11",
+        D_GLIBCXX_USE_CXX11_ABI,
+    ],
+    linkshared = 1,
+    deps = [
+        "@local_config_tf//:libtensorflow_framework",
+        "@local_config_tf//:tf_header_lib",
+    ],
+)

tensorflow_addons/custom_ops/text/cc/kernels/parse_time_kernel.cc

@@ -0,0 +1,103 @@
+/* Copyright 2019 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+#include <string>
+
+#include "absl/time/time.h"
+#include "tensorflow/core/framework/op_kernel.h"
+
+namespace tensorflow {
+namespace addons {
+
+using ::tensorflow::OpKernel;
+using ::tensorflow::OpKernelConstruction;
+using ::tensorflow::OpKernelContext;
+using ::tensorflow::Tensor;
+using ::tensorflow::tstring;
+
+enum OutputUnit {
+  SECOND = 1,
+  MILLISECOND = 2,
+  MICROSECOND = 3,
+  NANOSECOND = 4,
+};
+
+bool OutputUnitFromString(string output_unit_str, OutputUnit* output_unit) {
+  if (output_unit_str == "SECOND") {
+    *output_unit = SECOND;
+  } else if (output_unit_str == "MILLISECOND") {
+    *output_unit = MILLISECOND;
+  } else if (output_unit_str == "MICROSECOND") {
+    *output_unit = MICROSECOND;
+  } else if (output_unit_str == "NANOSECOND") {
+    *output_unit = NANOSECOND;
+  } else {
+    return false;
+  }
+  return true;
+}
+
+class ParseTimeOp : public OpKernel {
+ public:
+  explicit ParseTimeOp(OpKernelConstruction* context) : OpKernel(context) {
+    string output_unit_str;
+    OP_REQUIRES_OK(context, context->GetAttr("time_format", &time_format_));
+    OP_REQUIRES_OK(context, context->GetAttr("output_unit", &output_unit_str));
+    OP_REQUIRES(context, OutputUnitFromString(output_unit_str, &output_unit_),
+                errors::InvalidArgument("Invalid output unit"));
+  }
+
+  void Compute(OpKernelContext* context) override {
+    const Tensor& input_tensor = context->input(0);
+    auto input = input_tensor.flat<tstring>();
+
+    Tensor* output_tensor = nullptr;
+    OP_REQUIRES_OK(context, context->allocate_output(0, input_tensor.shape(),
+                                                     &output_tensor));
+
+    auto output_flat = output_tensor->flat<int64>();
+    const int n = input.size();
+    for (int i = 0; i < n; ++i) {
+      absl::Time time;
+      std::string err;
+      OP_REQUIRES(context, absl::ParseTime(time_format_, input(i), &time, &err),
+                  errors::InvalidArgument("Parse time failed: ", err));
+      switch (output_unit_) {
+        case SECOND:
+          output_flat(i) = absl::ToUnixSeconds(time);
+          break;
+        case MILLISECOND:
+          output_flat(i) = absl::ToUnixMillis(time);
+          break;
+        case MICROSECOND:
+          output_flat(i) = absl::ToUnixMicros(time);
+          break;
+        case NANOSECOND:
+          output_flat(i) = absl::ToUnixNanos(time);
+          break;
+      }
+    }
+  }
+
+ private:
+  std::string time_format_;
+  OutputUnit output_unit_;
+};
+
+REGISTER_KERNEL_BUILDER(Name("Addons>ParseTime").Device(tensorflow::DEVICE_CPU),
+                        ParseTimeOp);
+
+}  // end namespace addons
+}  // end namespace tensorflow

tensorflow_addons/custom_ops/text/cc/ops/parse_time_op.cc

@@ -0,0 +1,71 @@
+/* Copyright 2019 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+#include "tensorflow/core/framework/common_shape_fns.h"
+#include "tensorflow/core/framework/op.h"
+
+namespace tensorflow {
+namespace addons {
+REGISTER_OP("Addons>ParseTime")
+    .Input("time_string: string")
+    .Output("time_int64: int64")
+    .Attr("time_format: string")
+    .Attr("output_unit: {'SECOND', 'MILLISECOND', 'MICROSECOND', 'NANOSECOND'}")
+    .SetShapeFn(tensorflow::shape_inference::UnchangedShape)
+    .Doc(R"doc(
+Parse an input string according to the provided format string into a Unix time,
+the number of seconds / milliseconds / microseconds / nanoseconds elapsed since
+January 1, 1970 UTC.
+
+Uses strftime()-like formatting options, with the same extensions as
+FormatTime(), but with the exceptions that %E#S is interpreted as %E*S, and %E#f
+as %E*f.  %Ez and %E*z also accept the same inputs.
+
+%Y consumes as many numeric characters as it can, so the matching data should
+always be terminated with a non-numeric.  %E4Y always consumes exactly four
+characters, including any sign.
+
+Unspecified fields are taken from the default date and time of ...
+
+  "1970-01-01 00:00:00.0 +0000"
+
+For example, parsing a string of "15:45" (%H:%M) will return an Unix time that
+represents "1970-01-01 15:45:00.0 +0000".
+
+Note that ParseTime only heeds the fields year, month, day, hour, minute,
+(fractional) second, and UTC offset.  Other fields, like weekday (%a or %A),
+while parsed for syntactic validity, are ignored in the conversion.
+
+Date and time fields that are out-of-range will be treated as errors rather than
+normalizing them like `absl::CivilSecond` does.  For example, it is an error to
+parse the date "Oct 32, 2013" because 32 is out of range.
+
+A leap second of ":60" is normalized to ":00" of the following minute with
+fractional seconds discarded.  The following table shows how the given seconds
+and subseconds will be parsed:
+
+  "59.x" -> 59.x  // exact
+  "60.x" -> 00.0  // normalized
+  "00.x" -> 00.x  // exact
+
+time_string: the input time string to be parsed.
+time_format: the time format.
+time_int64: the number of seconds / milliseconds / microseconds / nanoseconds
+  elapsed since January 1, 1970 UTC.
+output_unit: the output unit of the parsed unix time. Can only be SECOND,
+  MILLISECOND, MICROSECOND, NANOSECOND.
+)doc");
+}  // end namespace addons
+}  // end namespace tensorflow

tensorflow_addons/text/BUILD

@@ -7,9 +7,11 @@ py_library(
     srcs = ([
         "__init__.py",
         "crf.py",
+        "parse_time_op.py",
         "skip_gram_ops.py",
     ]),
     data = [
+        "//tensorflow_addons/custom_ops/text:_parse_time_op.so",
         "//tensorflow_addons/custom_ops/text:_skip_gram_ops.so",
         "//tensorflow_addons/utils",
     ],
@@ -41,3 +43,16 @@ py_test(
         ":text",
     ],
 )
+
+py_test(
+    name = "parse_time_op_test",
+    size = "small",
+    srcs = [
+        "parse_time_op_test.py",
+    ],
+    main = "parse_time_op_test.py",
+    srcs_version = "PY2AND3",
+    deps = [
+        ":text",
+    ],
+)

tensorflow_addons/text/__init__.py

@@ -32,4 +32,7 @@
 
 # Skip Gram Sampling
 from tensorflow_addons.text.skip_gram_ops import skip_gram_sample
-from tensorflow_addons.text.skip_gram_ops import skip_gram_sample_with_text_vocab
+from tensorflow_addons.text.skip_gram_ops import skip_gram_sample_with_text_vocab
+
+# Parse Time
+from tensorflow_addons.text.parse_time_op import parse_time

tensorflow_addons/text/parse_time_op.py

@@ -0,0 +1,86 @@
+# Copyright 2019 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+"""Parse time ops."""
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+import tensorflow as tf
+
+from tensorflow_addons.utils.resource_loader import get_path_to_datafile
+
+_parse_time_op = tf.load_op_library(
+    get_path_to_datafile("custom_ops/text/_parse_time_op.so"))
+
+tf.no_gradient("Addons>ParseTime")
+
+
+def parse_time(time_string, time_format, output_unit):
+    """Parse an input string according to the provided format string into a
+    Unix time.
+
+    Parse an input string according to the provided format string into a Unix
+    time, the number of seconds / milliseconds / microseconds / nanoseconds
+    elapsed since January 1, 1970 UTC.
+
+    Uses strftime()-like formatting options, with the same extensions as
+    FormatTime(), but with the exceptions that %E#S is interpreted as %E*S, and
+    %E#f as %E*f.  %Ez and %E*z also accept the same inputs.
+
+    %Y consumes as many numeric characters as it can, so the matching
+    data should always be terminated with a non-numeric. %E4Y always
+    consumes exactly four characters, including any sign.
+
+    Unspecified fields are taken from the default date and time of ...
+
+      "1970-01-01 00:00:00.0 +0000"
+
+    For example, parsing a string of "15:45" (%H:%M) will return an
+    Unix time that represents "1970-01-01 15:45:00.0 +0000".
+
+    Note that ParseTime only heeds the fields year, month, day, hour,
+    minute, (fractional) second, and UTC offset.  Other fields, like
+    weekday (%a or %A), while parsed for syntactic validity, are
+    ignored in the conversion.
+
+    Date and time fields that are out-of-range will be treated as
+    errors rather than normalizing them like `absl::CivilSecond` does.
+    For example, it is an error to parse the date "Oct 32, 2013"
+    because 32 is out of range.
+
+    A leap second of ":60" is normalized to ":00" of the following
+    minute with fractional seconds discarded.  The following table
+    shows how the given seconds and subseconds will be parsed:
+
+      "59.x" -> 59.x  // exact
+      "60.x" -> 00.0  // normalized
+      "00.x" -> 00.x  // exact
+
+    Args:
+      time_string: The input time string to be parsed.
+      time_format: The time format.
+      output_unit: The output unit of the parsed unix time. Can only be SECOND,
+        MILLISECOND, MICROSECOND, NANOSECOND.
+
+    Returns:
+      the number of seconds / milliseconds / microseconds / nanoseconds elapsed
+        since January 1, 1970 UTC.
+
+    Raises:
+      ValueError: If `output_unit` is not a valid value,
+        if parsing `time_string` according to `time_format` failed.
+    """
+    return _parse_time_op.addons_parse_time(time_string, time_format,
+                                            output_unit)